In today's computing environment, user's satisfaction is largely dependent on the continuous non-stop service availability. However, system upgrade is inevitable and may even need to be very frequent, for example, for patches and intra-release upgrades that provide minor fixes and feature additions and does not require data model changes.
Traditional upgrades in a computing environment such Infrastructure as a Service (IaaS) requires a designated maintenance window, during which the disruption of service may occur. This disruption of service may have a significant impact on the user or customer who would like to access the service, and may result in lower user or customer satisfaction to the service. On one hand, if the service of the computing environment is offered globally, choosing an appropriate maintenance window is difficult and may be dependent on geographical locations. Considering that the system patches or intra-release upgrades may be as frequent as weekly, the negative impact of traditional maintenance window approach can be significant.